“Networks, Societies, Spheres: Reflections of an Actor-­‐
network Theorist”* Keynote speech for the INTERNATIONAL SEMINAR ON NETWORK
THEORY: NETWORK MULTIDIMENSIONALITY IN THE DIGITAL AGE
19th February 2010
Annenberg School for Communication and Journalism
Los Angeles
Bruno Latour, Sciences Po
In one way or another we are all in this room responsible for having given to
the notion of networks an immense, and some could say, an hegemonic extension.
Either because some of you have created the hardware and software infrastructure
that has added digital networks to the already existing water, sewage, road, rail,
telegraph, and telephone networks, or because others, through media studies,
sociology, history, political sciences, and even philosophy and brain science, have
tried to capture what is so original in the new networky world generated by those new
socio-technical assemblages. The reason why I have welcomed the kind invitation of
Professor Manuel Castells, is that, because of the very extension of network (as a thing
of the world as well as a concept), the time has come to check what it really means
and maybe to shift somewhat its ambition and modify its real import. When a notion
has become enshrined into a work of art like Cameron’s AVATAR with the planet
Pandora itself sprouting its billions of web by connections and the very notion of
communication among the Na’vis and their creatures being materialized by a real
plug-in of hair, tails and manes, it is time to stop and ask: “what have we done?”
This paper is the result of a collective work carried out at the médialab of
Sciences Po in collaboration with Dominique Boullier, Paul Girard, and Tommaso
Venturini. It has been funded under the European project MACOSPOL.
*
121-CASTELLS-NETWORKS
2
I hope you accept that I include myself into this highly professional “we,” not
because I have contributed anything to the deployment of digital network, not
because I have studied the extension of the various socio-technical systems of
information and communication, but because, for about thirty years now, I have
found in the notion of network a powerful way of rephrasing basic issues of social
theory, epistemology and philosophy. Consider me, then, as a fellow traveler of the
various network revolutions (if revolution is the word).
I am well aware that there is nothing more perilous than an after dinner
keynote since the speaker is supposed to entertain the audience with witty anecdotes. I
have chosen another tack: to make it more memorable, I decided to make it as
earnest and as boring as possible…
In its simplest but also in its deepest sense, the notion of network is of use
whenever action is to be redistributed.
This is well known in my field of science and technology studies. Take any
object: at first, it looks contained within itself with well delineated edges and limits;
then something happens, a strike, an accident, a catastrophe, and suddenly you
discover swarms of entities that seem to have been there all along but were not visible
before and that appear in retrospect necessary for its sustenance. You thought the
Columbia shuttle was an object ready to fly in the sky, and then suddenly, after the
dramatic 2002 explosion, you realize that it needed NASA and its complex
organizational body to fly safely in the sky —here is the hall where the disjointed parts
have been assembled for the task force to inquire into what went wrong (Figure 1).
The action of flying a technical object has been redistributed throughout a highly
composite network where bureaucratic routines are just as important as equations
and material resistance. Yes, it is a strange space that of a shuttle that is just as much
in the sky as inside NASA, but that’s precisely the space—hard to describe and even
harder to draw—that has been made visible by the deployment of networks in my
sense of the word.
121-CASTELLS-NETWORKS
3
Figure 1. The shuttle flies in space and in the NASA corporate structure.
The same transformation has happened in epistemology. I borrow the following
example from the great historian of science, Simon Schaffer: you believed Newton’s
PRINCIPIA MATHEMATICA (Figure 2) had been written in the complete isolation of a
totally bodiless mind, and you suddenly discover that Newton was more like a spider
in the center of a huge web that covered every possible type of witness carrying and
sending information back and forth—these are maps of the world with the flags
representing the precise spots from which Newton requested precise information
(Figure 3). Here again, the notion of networks points to a transformation in the way
action is located and allocated. Here again, what was invisible becomes visible, what
had seemed self contained is now widely redistributed. Newton reaches the stars
because he is also the center of a vast empire of information. Not because of an
accident, as in the shuttle example, but, interestingly enough, because of the wide
transformation in our worldviews that the very notion of network has introduced into
the new history of science. The search for the production of object and of objectivity
is totally transformed now that they are portrayed simultaneously in the world and
inside their networks of production. This is the contribution of my field, science and
technology studies, of which I am the most proud.
121-CASTELLS-NETWORKS
Figure 2. The first English translation of Mathematical Principles (1729).
Figure 3. Source of data for Mathematical Principles. Source: Simon Schaffer, The
Information Order of Isaac Newton's Principia Mathematica, Salvia Smaskrifter, Upsalla
(2008).
4
121-CASTELLS-NETWORKS
5
You see that I take the word network not simply to designate things in the world
that have the shape of a net (in contrast, let’s say, to juxtaposed domains, to surfaces
delineated by borders, to impenetrable volumes), but mainly to designate a mode of
inquiry that learns to list, at the occasion of a trial, the unexpected beings necessary for
any entity to exist. A network, in this second meaning of the word, is more like what
you record through a Geiger counter that clicks every time a new element invisible
before has been made visible to the inquirer.
To put it at its most philosophical level (not a thing to do, I know, after dinner
at night…), I’d say that network is defined by the series of little jolts that allow the
inquirer to register around any given substance the vast deployment of its attributes.
Or, rather, what takes any substance that had seemed at first self contained (that’s what
the word means after all) and transforms it into what it needs to subsist through a
complex ecology of tributaries, allies, accomplices, and helpers (I chose the word
ecology on purpose as will be clear later). The shuttle Columbia was not an object
whose substance could be defined, but an array of conditions so unexpected that the
lack of one of them (a bureaucratic routine) was enough to destroy the machine;
Newton’s sublime system was not a self contained substance, but a vast empire of
information necessary for the system to subsist and expand. Whenever a network is
deployed, a substance is transformed from an object into a thing, or to use my terms,
from a matter of fact to a matter of concerns. If we still want to use the term of
“network revolution,” it is in that sense, I believe, that it can be said to be a revolution
and clearly a political one.
The ability of the notion of network to follow this strange movement that goes
from substance to attributes and back, as if you could follow the movement of a fan
that one could chose to close or to deploy, is at the heart of this rather (in)famous
social theory known as actor-network theory, abbreviated in the felicitous acronym
ANT. By the way, I am sorry to say that what I mean by actor-network bears no
relation with the same term in Barabassi’s LINKED, by which he means the league or
the union of real actors from nearby Hollywood! No, alas, mine is a purely conceptual
term that means that whenever you wish to define an entity (an agent, an actant, an
actor) you have to deploy its attributes, that is, its network. To try to follow an actornetwork is a bit like defining a wave-corpuscle in the 1930s: any entity can be seized
either as an actor (a corpuscle) or as a network (a wave). It is in this complete
reversibility—an actor is nothing but a network, except that a network is nothing but
actors—that resides the main originality of this theory. Here again, network is the
concept that helps you redistribute and reallocate action.
Now, this is where things become complicated and where the digital expansion
given to information techniques is going to have huge and fascinating effects.
121-CASTELLS-NETWORKS
6
But before I review some of those effects, I’d like to introduce you to the work of
the artist Tomas Saraceno because he has offered a powerful view of how networks,
spheres and tensors could actually fit together (Figure 4). As you may know, one of the
criticisms often made about networks (particularly by Peter Sloterdijk) is that they are
extremely poor metaphors since they remain entirely made of nodes and edges to
which is often added some conveniently drawn potato-like circles (I will come back to
this impoverished visual vocabulary later). To say that something is a network is about
as appealing as to say that someone will, from now on, eat only peas and green beans,
or that you are condemned to reside in airport corridors: great for traveling,
commuting, and connecting, but not to live. Visually there is something deeply wrong
in the way we represent networks since we are never able to use them to draw
enclosed and habitable spaces and envelopes.
121-CASTELLS-NETWORKS
7
Figure 4. Art installation by Tomas Saraceno, Venice Biennale.Photo BL
Well, the great virtue of Saraceno’s installation (this one in the latest 2009
Venice Biennale) is that he has managed to obtain comfortable and enclosed spherical
sites which are nonetheless entirely made of networks. The trick, as you can see, is in
changing the density of connections until a net ends up being undistinguishable from
a cloth. And the work of art is even better because neither spheres nor nets are
actually the real physical thing, which is made of elastic tensors carefully arrayed and
fixed on the walls. A beautiful case of action being redistributed, since visitors are able
to check for themselves (when there is no guard around, that is), by pushing or pulling
a tensor, what else is moving in the whole array. Like his mentor Olafur Eliasson,
Saraceno is one of those artists who is exploring, often more daringly than social
theorists, visual possibilities where self-contained substances are captured with their
attributes fully deployed. This is why they are rightly called “ecological artists.” Is not
ecology anything but the deployment of all the attributes necessary for any selfcontained entity to subsist? To be self-contained—that is to be an actor—and to be
thoroughly dependent—that is to be a network—is to say twice the same thing. As
Gabriel Tarde (a character to which I will come back) said: the reason why this is not
common sense is because philosophers have been carried out by the verb to be and its
problem of identity and not by the verb to have and the range of its properties and
avidities. But the web is changing all of that and fast: “to have” (friends, relations,
profiles…) is quickly becoming a stronger definition of oneself than “to be.”
121-CASTELLS-NETWORKS
8
I am sorry to insist so much on the conceptual aspect of network, but this is
because I have found this notion useful long before it gained its new incarnation in
real life-size nets, webs, and Gaia-like planets (like Earth or Pandora). What I have
always found great in the metaphor of the net is that it is then easy to insist on its
fragility, the empty spot it leaves around (a net is made first of all of empty space), the
subversion it introduces in the notion of distance (the adjectives “close” and “far” are
made dependant on the presence of conduits, bridges, and hubs), but above all, what
it does with universality: the area “covered” by any network is “universal” but just as
long and just where there are enough antennas, relays, repeaters, and so on, to sustain
the activation of any work. Thanks to the notion of networks, universality is now fully
localizable. In network, it’s the work that is becoming foregrounded, and this is why
some suggest using the word worknet instead.
But what I like most in the new networks is that the expansion of digitality has
enormously increased the material dimension of networks: the more digital, the less
virtual and the more material a given activity becomes. Nowadays, everyone knows that
there is no GPS without three satellites; collective games without fast connections;
drones in Pakistan without headquarters in Tampa, Florida; bank panic without
Reuters screens; and so on. When Proust could read a novel alone hidden in the
shack of Combray, it was possible to say that his imaginary mental world was virtual,
but we can’t say that of our kids who have to hook up their modems, buy game
stations, swap disks and pay their server for a faster connection with our credit card.
Young Marcel could build castles in Spain (pie in the sky?) for nothing, now he would
have to buy real estate on Second Life with hard Linden dollars. When Harold
Garfinkel described the skills necessary to “pass” as a member of a society, you could
say it was a totally intangible social phenomenon that could be only qualitatively
described, but not today when every detail of your avatars on the web can be
counted, dated, weighed, and measured. Then you know that everything that before
had melted into air has become fully incarnated. Go tell Google engineers that their
vast arrays of servers are just virtual! This is probably the greatest and yet the least
celebrated feat of your collective work, Ladies and Gentlemen: to have rendered fully
visible what is needed to think and to imagine and to trust; to have taught all of us
that those cognitive competences are now paid in hard won bits and bytes—and have
become, for that reason, fully describable.
To sum up: whenever an action is conceived as networky, it has to pay the full
prize of its extension, it’s composed mainly of voids, it can be interrupted, it is fully
dependent on its material conditions, it cannot just expand everywhere for free (its
universality is fully local). Networks are a great way to get rid of phantoms such as
nature, society, or power, notions that before, were able to expand mysteriously
121-CASTELLS-NETWORKS
9
everywhere at no cost. As the study of metrology, standards, empires, has shown so
well, smooth continuity is the hardest thing to get.
I hope you now understand that if we accept to talk about a network revolution,
it is because of the coincidence between the conceptual notion of network (action is
radically redistributed) and the rematerialization allowed by digital techniques. As a
sociologist of sort, I have been especially interested in what this revolution does to
social theory. And what it does is truly amazing: it dissolves entirely the individual
versus society conundrum that has kept social theorists and political scientists busy for
the last two hundred years. To sum up a long argument: we have the social theory of
our datascape. If you change this datascape, you have to change the social theory.
Why do we think that they are individuals who are “in” a society? Because of a
discontinuity in the available data. When we gather statistics—and this is what social
theorists have done for the last hundred and fifty years when they were not doing
qualitative field work—the sheer difficulty of getting the data means that you are
going to focus on the individual as little as possible in order to get as quickly as
possible at the aggregates. Inevitably, you are going to begin to grant to those
aggregates some sort of existence by themselves. This is where the notion of society is
generated, a special way to grasp collective phenomena that Durkheim has defined by
the word sui generis or that you find just as well in the tired old cliché that “the whole is
superior to the sum of its parts.” Once you are there, social theory is finished,
sterilized for a whole century: you have parts, and you have a whole. And then the
only remaining question is to find a possible solution to combine or reconcile the parts
with the whole, a question which, as you know so well in this benighted country of
yours, is not an academic one since it throws people in the street —as the Tea Party
movement demonstrates vividly enough. Self-contained individuals fight for a place in
the self-contained society.
My claim, or rather ANT’s claim and that of the revisited tradition dating from
the great French sociologist, Gabriel Tarde, at the turn of the 19thcentury, is that the
very idea of individual and of society is simply an artifact of the rudimentary way data
are accumulated (Figures 5 and 6). The sheer multiplication of digital data has
rendered collective existence (I don’t use the adjective social anymore) traceable in an
entirely different way than before. Why? Because of the very techniques that you,
Ladies and Gentlemen, have brought to the world.
121-CASTELLS-NETWORKS
Figure 5. The Tarde/Durkheim 1904 debate replayed in Paris. http://www.brunolatour.fr/expositions/debat_tarde_durkheim.html
Figure 6. Map of the blogosphere built around French politics, by RTGI now
Linkfluence http://fr.linkfluence.net/
10
121-CASTELLS-NETWORKS
11
There is nothing easier now than to navigate back and forth from an individual
profile to an aggregate of hundreds and thousands of profiles. But the whole novelty is
precisely in the possibility of going back and forth. Before, in the old days of traditional
statistics, this was exactly the steps that one could not easily retrace: of course you
could in principle go back from a compiled questionnaire to the individual tick on the
form, but the guy who had ticked the form has long disappeared —no inquirer could
trace it back. Hence a discontinuity, a disjunction introduced in the traceability of the
associations. The less you can go back to the individual transaction, the more
tempting it is to give to the aggregate a substantial reality. But today, every one of us,
because of the navigational movement made possible through the datascapes on the
screen, is able to reintroduce a continuity from individual contribution to the
aggregates in a much more smoother way than before. (The experience is possible
only in front of the screen; it’s much harder to keep this focus on a piece of paper, and
this is why it is not described so much).
And what is the result of this new habit of navigating back and forth through
datascapes without stopping at either of their two end points? Well, the two extreme
points at which the whole of social theory had solidly fastened their Big Questions —
that is, the individual versus the society, who should take precedence, and how power
is exerted from one to the other, and so on and so forth— begins to lose their
undisputed privilege and even, after a while, vanish away. Instead of THE individual
versus society Problem, we are now faced with the multiple and fully reversible
combinations of highly complex individual constituents and multiple and fully
reversible aggregates. The center stage is now occupied by the navigational tools.
I believe it is the unique and unexpected combination of, first, the datascapes, second,
the navigation skills acquired on the screen, and, third, actor-network theory, that has
totally redistributed the classical arguments of a society made of individuals. (It is not
a small paradox that this alternative theory of the social had been anticipated a
century ago by Gabriel Tarde, a keen connoisseur of contemporary statistics who had
detected immediately in the project of his young colleague Emile Durkheim the
danger of introducing much too fast a discontinuity between two levels: that of
individual psychology and that of a sui generis society—and the perversity of their
debate is that it is Durkheim, the one who invented the two levels principle, who has
been able to persuade his readers that it was Tarde who occupied one of the two
positions, that of individual psychology, whereas Tarde had, on the contrary, denied
that there were two levels and tried to bypass entirely the two end points of
individuals and society. Needless to say Durkheim won and Tarde lost, until, that is,
the web came in to vindicate him by offering at last, if I dare say so, a nonindividualistic grasp on the individual!).
121-CASTELLS-NETWORKS
12
The reasonable thing for me to do, so late at night, would be to stop there and
to crack a few jokes to help you digest your dinner before having a sip of cognac. And
yet, I cannot resist the temptation to explore further with you some of the odd
consequences of this redistribution of action allowed by the concept of network
combined with the development of digital datascapes. I am afraid cognac will wait a
bit.
Even though it seems commonsense to say that the whole is superior to the parts, a
minute of reflection is enough to realize that this is due to the introduction of the
discontinuity in data collection I mentioned earlier: you notice individuals reduced to
very few properties walking or working in downtown Los Angeles; then you look at
the huge skyscrapers that tower above them; and then it seems reasonable to say that
“the whole is superior to the parts,” or that there emerge out of individual interactions
many things that the individual had not anticipated. . Possibly. But this does not mean
that at some point the action of individuals has been taken over ex abrupto by some sui
generis entity that could be called Los Angeles society. That is precisely the point that
Tarde always objected to with Durkheim: we know from firsthand experience that
this never happens. It does not mean that there is no society and only individuals (an
accusation leveled at Tarde by Durkheim), it means that the two notions are the two
faces of the same coin and this coin has no more currency any more than a French
franc.
To believe in the existence either of individual or of society is simply a way to
say that we have been deprived of information on the individuals we started with; that
we have little knowledge about their interactions; that we have lost the precise conduits
through which what we call “the whole” actually circulates. In effect, we have
jettisoned the goal of understanding what the collective existence is all about. Is it not
strange to imagine a science of society making sure that its main phenomenon will be
forever rendered impossible to detect and to document?
Now suppose that we benefit, thanks to digital techniques, from a vast range of
information about individuals. Let us be careful here: by individual I don’t mean the
individual atoms deprived of most of their properties and rendered fully
interchangeable before they enter into “interactions”. Instead of those atomic
individuals of the past, we now possess individuals for which we are allowed to
assemble profiles made of long lists of properties. Nothing is more common on the web
than this explosion of profiles willingly or unwittingly accumulated, stored, treated,
and visualized. Until the digital techniques of capture and storage, many fields of
social sciences, as you know, had been divided between qualitative and quantitative
research (I am myself a qualitativist having done mostly field work). But individual
profiles begin to seriously blur the distinction between the two sets of skills. Contrary
121-CASTELLS-NETWORKS
13
to common wisdom, and exactly as predicted by Tarde, the more you individualize the more
you can quantify —or else we have to find another name than quantification to describe
the phenomenon (is quali-quantitative a possible term?).
Why is it that the substitution of long and complex individual profiles to that
atomic individual generates such a difference in the actor/system conundrum?
Because when we begin to gather profiles, the very notion of interaction begins to
wobble. The reason is that a given individual will be defined by the list of other
individuals necessary for its subsistence. This is the reversibility of actor and network
mentioned earlier, or that of substance and attributes. Every individual is part of a
matrix whose line and columns are made of the others as well. To take the example
not of downtown Los Angeles but of the recently rediscovered METROPOLIS, it would
be easy to build a database where Freder Fredersen is defined as son of the Joh
Fredersen, loves Maria, befriends worker n°1255, etc., and then to ask any good
social network software to automatically permutate for you the positions so that Joh
Fredersen will in turn also be defined as the father of Freder, the enemy of Rotwang
and of Maria, etc., etc.
If we pursue this thought experiment we realize that we have already solved (or
rather dissolved) one of the classical problems of social theory: the reason why people
said that interactions create phenomena superior to the individual social atoms, is
because they had first defined the atoms as self contained entities deprived of all the other
entities necessary for their subsistence (they had failed to see actors as actor-networks).
No wonder that then, when entering any interaction, those simplified and castrated
atoms had produced unintended consequences: too little was known about them in
the first place! Strictly speaking, it is not true that there are interactions between
individuals. Individual action is much too distributed to be defined in terms of
interaction. This is one of the first strange consequences of taking seriously the notion
of actor-network.
But the second consequence of gathering so much information about individual
profiles is even stranger: the very notion of the whole begins to be deeply modified.
What is a collective phenomenon once you deploy all the information you have about
individual associations? It is certainly not something superior to the web they form by
sharing their profiles. What is it then? Probably something inferior, something smaller
than the parts. This is what Tarde always objected to with Durkheim: the whole is
necessarily less complex than the individual who makes it possible, provided, that is,
you accept not to reduce individuals to self-contained atomic entities but let them
deploy the full range of their associates—which means of course that you need to
have a lot of information about their profiles.
This argument seems bizarre only because we are used to the three usual
metaphors that have been developed over the course of the centuries to talk about
121-CASTELLS-NETWORKS
14
collective phenomena: (a) a society overarching individuals—the organicist metaphor;
(b) an invisible hand producing optimum out of simple minded atomic calculators—
the economic metaphor, or (c) an emerging structure—the auto-organization
metaphor. All of those start with atomic individuals and imagine a second level where
the collective phenomenon takes over. But it might be the time to imagine other
metaphors where there is only one level, where the parts are actually bigger than the
whole and where a phenomenon can be said to be collective without being superior to
individuals. A better metaphor would be the one that would rely for instance on the
ways in which standards circulate through the net, or fashion, buzz, epidemics, that is,
just the sort of things that are now easy to detect, to follow and to visualize with the
new digital tools made available (Figure 6).
Figure 6. Network datascapes illustrating phenomena in which the collective is not
superior to the individuals. http://www.visualcomplexity.com/vc/
You are going to tell me that this kind of information on the building of
downtown Los Angeles that I took as my example is totally inaccessible so that my
thought experiment is just that, a thought, not an experiment. Maybe, but it is not the
same thing to say that because of a lack of information we speak as if there was a
whole superior to the parts, or to say that the great problem of social theory is to
“reconcile the actor and the system”. What is shut close with the second formulation
is wide open with the first. And I could add that there exist many sites where we do
have this information for instance in the artificial worlds of SIMCITY. Or, even more
tellingly, in the many efforts of many interesting radical architects, planners, and
121-CASTELLS-NETWORKS
15
builders to devise digital platforms to resolve the question of collective or participatory
design.
There is something always fishy and I believe deeply wrong in the idea of a
whole superior to its parts. I have always the feeling that we have not moved much
from Menenius Agrippa’s famous simile of “the Members and the Belly”. Remember
CORIOLANUS. If you accept the notion of organism as something different or superior
or even emerging, you lose what an organization is (and I would add you ruin the
possibility of doing politics). A phenomenon may be collective without being social.
The reason I insist on this far-fetched argument is that it opens a much more
interesting collaboration between sociologists and for instance biologists fighting
against the equally misleading notion of an organism (organizations and organisms
share the same paralyzing social and political theory) or between sociologists and
neuroscientists. Since there is no conductor, nor homunculus, nor sui generis society
anywhere, we might be able to collaborate more effectively by following the right
conduits through that which appeared before as a whole above parts but is actually a
part, primus inter pares, so to speak, that traverses through the parts. The problem is the
same in a brain, in a body, in a city. Yes, networks are everywhere but not quite in
Barabassi’s generalization of a world made of links. Rather in the neo-Leibnitizian
meaning of the word that Tarde had resurrected under the name “monads.”
There is something actually very bizarre in the attempts to apply models
borrowed from natural sciences to social phenomena. Too often, physicists or
biologists, try to make individual human atoms just as simple minded as atoms in
physics or ants in entomology. Now, I have nothing against models (in my médialab
we are actually trying to model Tarde’s idea of a whole smaller than the part society).
But is it not strange to claim to imitate the natural sciences while doing just the
opposite? What is so striking in human societies is how much information is available
on individual profiles; so it is a bit silly to say nonetheless that we should start with
interchangeable atoms. A reasonable and apparently fully scientific way would be to seize
the opportunity offered by the mass of information now available. And yet, what is
done instead? The humans (on which masses of information are available) are treated
as atomic morons on which as little as possible is known, by endowing them with as
few rules of behavior as possible, so that they generate through their “interaction” (a
loaded term as we just saw) as complex a structure as possible. And all of that in the
name of imitating for instance the study of ants (I mean the real ones not ANT!).
But when entomologists made the startling discoveries that they could explain
the building of elaborate structures such as the anthill without relying on any notion of
superorganism, this is exactly what should be done with human societies. With this
important difference that humans dispose of billions of neurons and not tens or
121-CASTELLS-NETWORKS
16
hundreds of thousands like social insects. So, what does it mean to really imitate the
natural sciences: is it to start from humans with billions of neurons about which we
possess elaborate profiles in huge databases and then strip them bare so that they end
up looking like ants? Or is it to do exactly as it has been done with ants, that is get
entirely rid of the notion of superorganism and even of that of two levels, and to try to
see how those monads manage to build elaborate structures without ever relying on a
whole superior to the parts?
The true digital revolution in social theory is to open a way whereby it is
possible to study the individuals and their aggregates without relying at any point on
two levels, without accepting any discontinuity where the individual action disappears
mysteriously into a sui generis structure. I really believe, Ladies and Gentlemen, that if
we succeed in doing this we will achieve for human societies discoveries just as
revolutionary as what has been done with insect societies—and without in the least
looking reductionist since we will not have to commit the rather silly mistake of
discarding all the available information to limit humans to ants or atoms just because
physicists and biologists like to have masses of interchangeable elements for their
models. Why not trying to move from complexity—the parts—to simplicity—the
whole—instead of doing the opposite? Since the information is here why not use it?
Actually, there are good reasons for not using it, and I will end this lecture with
two of those, just in case you find yourself too excited about the prospect I am offering
you…
The first is the one I alluded to earlier: the mass of data available is accessible
through an incredibly poor visual datascape. Actually, the word datascape is somewhat
of a misnomer. It is not a pleasant landscape, but rather like watching lines and lines
of barbed wire. How tiring it is to ponder click after clicks all those nodes and all those
edges and all those potato shaped lines. When Tarde predicted, a century ago, that
when statistics would be really good, social phenomena would be as pretty and easy to
look at as the flight of a swallow, how disappointed he would be to look at the anemic
spaces of the web. It is called “visual complexity” but it is actually not complex at all,
nothing at least like the sight of a flying swallow. Poor and boring and even when
agitated by flashy and sexy moving gadgets it is just as informative as the reading of
tea leaves. I don’t want to sound too impolite, Ladies and Gentlemen, but I think you
could do much better! The whole world is expecting from you visual instruments
which are at the level of the extraordinary transformations brought about by the
traceability of collective phenomena and compatible with our very efficient visual
skills. It took about eighty years for statistics to become a vocabulary for doing social
sciences. We should be able to speed up the time necessary to transform the mass of
quali-quantitative data into agreed upon and comfortable looking datascapes. Which
121-CASTELLS-NETWORKS
17
of course means that we should be able to solve the question of compounding masses
of individual profiles in a fully reversible way, that is exactly what traditional statistics
have not been able to do.
Now, the difficulties of realizing those major transformations in visualization
and computing are compounded by another even more formidable challenge. We
should be able to navigate through datascapes which are not only visually coherent
but which are also able to follow controversies. This is what I call solving the
Lippmannian problem (that problem that Walter Lippmann addressed so well in his
masterpiece THE PHANTOM PUBLIC). The social theory question of bypassing the
individual versus society is exactly paralleled by the epistemological question of obtaining
authority while bypassing the distinction between rational and irrational voices. It is
actually twice the same problems and this is why their connection is at the core of
ANT. The recent climategate fracas is a good case in point: how do you map the
controversy around the evolution of climate without resorting either to conspiracy
theories or to the positivist narrative that Earth’s climate speaks directly to the GIEC’s
scientists much like Eywa speak to the Na’vis? Two types of fundamentalisms which
are fiercely opposed because they resemble themselves so much: a self-contained
authority that would need no network of attributes to be sustained? So much talk
about sustainable development, and so little attention given to what makes argument
sustainable!
I have been directly engaged in this last question through the creation of my
school, the tiny tiny médialab, the fifteen year development of a course called
“cartography of scientific controversies,” and a now finished European project call
MACOSPOL (MApping COntroversies on Science for POlitics) to try to develop a
platform for making comfortable for scientists and users of scientific data the
navigation through controversial datascapes. If I wanted to dramatize somewhat the
general problem we all face, I would say that what we have to do is to reinvent the
newspaper in a completely new form (this is why Lippman’s wisdom is so important).
If it is true, as many historians have shown, that there is a direct link between the
invention of the newspaper and the possibility for citizens to articulate political
opinions, and if it is true that the old newspaper appears retrospectively as a platform
connecting heterogeneous data, then it is extremely urgent to reinvent a platform no
longer on paper but in the newly rematerialized world of digital datascapes. Digital
democracy has generated a lot of hype, but I believe, as many of you here, that its
true development is still to come and that it will be necessary to invest also, in no
small part, in the theoretical import of the notion of network as this conference
proposes to do. When Lippmann said the public is a phantom, this was not a way to
say it does not exist, but on the contrary a plea—and a somewhat desperate plea—to
make it appear through the invention of the right tools. It is only because of the
121-CASTELLS-NETWORKS
18
importance of the task at hand and of the seriousness of the challenge that I have
taken the liberty tonight of submitting to you those remarks on the theory of network,
fully conscious that you know infinitely more than me on those various challenges but
equally conscious that not one of them can be met without a collaboration between
many various fields including philosophy. Thank you very much Ladies and
Gentlemen for your patience.